feat(g4f): Major provider updates and new model support

[kqlio67]

Core changes:

• Added new models: o1-preview, o1-mini, llama-3.2-90b, codellama-34b, mixtral-7b, mistral-tiny, claude-3.5-haiku, qwen-2.5-72b, flux-dev, dalle-3, recraft-v3
• Added new providers: Copilot, LLMPlayground, AmigoChat
• Added new base providers: Gryphe, databricks, anthracite-org, ai21, llmplayground.net

Provider improvements:

• Enhanced DDG and Copilot with better conversation handling
• Fixed system prompts in ChatGptEs and Blackbox (#2393), (https://github.com/xtekky/gpt4free/issues/2433#issue-2695811324)
• Improved message processing in Airforce (#2360)
• Updated Liaobots and DeepInfra with latest models
• Added Qwen2.5 and Nemotron support to multiple providers

Maintenance:

• Moved PollinationsAI to needs_auth/
• Moved AIUncensored to not_working/
• Removed deprecated Allyfy provider
• Updated model mappings and provider configurations

Breaking changes:

• Removed several outdated models from Liaobots

[TheFirstNoob]

Hi! I still think airforce and llm providers should be divided into two providers, and not replace each other. Based on tests from version 0.3.3.4 and any subsequent ones, if you use llm settings, then there are much fewer working text-models. If you use airforce on version 0.3.3.4, then there are much more working text-models, although they are not specified there.

Also, Chatgpt4Online and FreeNetfly are not removed from the model.py. These providers are removed, but we still call them in this file. They should be remove.

[github-actions[bot]]

Pull Request Review: feat(g4f): Major provider updates and new model support

Summary

Thank you for your contribution to the project! This pull request introduces significant updates including new models and providers, enhancements to existing providers, and necessary maintenance changes.

Core Changes

• New Models Added:

  • o1-preview, o1-mini, llama-3.2-90b, codellama-34b, mixtral-7b, mistral-tiny, claude-3.5-haiku, qwen-2.5-72b, flux-dev, dalle-3, recraft-v3

• New Providers Added:

  • Copilot, LLMPlayground, AmigoChat

• New Base Providers:

  • Gryphe, databricks, anthracite-org, ai21, llmplayground.net

Provider Improvements

• Enhanced conversation handling in DDG and Copilot.
• Fixed system prompts in ChatGptEs and Blackbox.
• Improved message processing in LLMPlayground.
• Updated Liaobots and DeepInfra with the latest models.
• Added Qwen2.5 and Nemotron support to multiple providers.

Maintenance

• Moved PollinationsAI to needs_auth/.
• Moved AIUncensored to not_working/.
• Removed deprecated Allyfy provider.
• Renamed Airforce to LLMPlayground.
• Updated model mappings and provider configurations.

Breaking Changes

• Removed several outdated models from Liaobots.
• Replaced Airforce with LLMPlayground in the provider list.

Code Quality

The code changes are well-structured and the addition of new models and providers seems to follow the existing conventions. The use of constants for models and personas enhances maintainability.

Suggestions

• Consider adding unit tests for the new models and providers to ensure functionality is as expected.
• Documentation updates would be beneficial to reflect the new providers and models added.

Once again, thank you for your contribution! Your efforts help improve the project significantly.

[github-actions[bot]]

Review of PR: feat(g4f): Major provider updates and new model support

Core changes:

• Added new models:
  • o1-preview, o1-mini, llama-3.2-90b, codellama-34b, mixtral-7b, mistral-tiny, claude-3.5-haiku, qwen-2.5-72b, flux-dev, dalle-3, recraft-v3
  • meta-llama/Meta-Llama-3.2-90B-Vision-Instruct-Turbo, codellama/CodeLlama-34b-Instruct-hf, meta-llama/Llama-3.2-90B-Vision-Instruct-Turbo, mistralai/Mistral-7B-Instruct-v0.3, mistralai/mistral-tiny, deepseek-ai/deepseek-llm-67b-chat, databricks/dbrx-instruct, anthracite-org/magnum-72b, ai21/jamba-1-5-mini
• Added new providers:
  • Copilot, LLMPlayground, AmigoChat
• Provider improvements:
  • Enhanced DDG and Copilot with better conversation handling.
  • Added Qwen2.5 and Nemotron support to multiple providers.
  • Fixed system prompts in ChatGptEs and Blackbox.
  • Improved message processing in LLMPlayground.
• Maintenance:
  • Moved PollinationsAI to needs_auth/.
  • Moved AIUncensored to not_working/.
  • Removed deprecated Allyfy provider.
  • Renamed Airforce to LLMPlayground.
  • Updated model mappings and provider configurations.

Breaking changes:

• Removed several outdated models from Liaobots.
• Replaced Airforce with LLMPlayground in provider list.

Review:

Thank you for the extensive updates and the addition of new models and providers. This pull request is very comprehensive and introduces significant improvements to the project. The detailed changes and additions will enhance the functionality and usability of the tool significantly.

Recommendations:

1. Ensure that the new models and providers are thoroughly tested in different environments to validate their performance and compatibility.
2. Consider adding more detailed documentation for the new models and providers to help users understand their capabilities and usage.
3. Conduct thorough testing to ensure no existing features are broken by these updates.

Overall, this pull request is well-executed and will bring valuable enhancements to the project. Keep up the good work!

Thank you for contributing to the project!

[github-actions[bot]]

g4f copilot review

Thanks for the significant contribution to the gpt4free project! This pull request introduces a substantial number of improvements, adding new models and providers and enhancing existing ones. The changes are well-documented, and the commit messages are clear and concise.

Here are some observations and suggestions:

Positive Aspects:

• Comprehensive Updates: The addition of numerous models and providers significantly expands the capabilities of gpt4free.
• Improved Providers: The enhancements to existing providers, such as improved conversation handling for DDG and Copilot, are valuable additions.
• Clear Documentation: The pull request description provides a thorough overview of the changes, including a helpful list of breaking changes.
• Issue Resolution: Addressing existing issues (#2393, #2360) demonstrates proactive maintenance and problem-solving.
• Refactoring: Moving providers to needs_auth/ and not_working/ and removing deprecated providers is good housekeeping.
• Extensive Testing: The sheer number of updates suggests a significant amount of testing was performed.

Areas for Improvement/Questions:

• AmigoChat Model Mapping: The AmigoChat provider now uses a large dictionary MODELS for model mapping. While functional, consider if a more structured approach (e.g., a class or separate configuration file) might be more maintainable as the number of models grows.
• Error Handling: The AmigoChat provider mentions balance checks. It might be beneficial to add more robust error handling and informative error messages for these situations.
• Airforce (LLM Playground) Chunking: The message chunking implemented in Airforce seems potentially useful in other providers. Consider if this logic could be abstracted into a reusable helper function for broader application.
• Testing Strategy: While extensive testing is implied, including some specific details in the PR description about the testing approach (e.g., unit tests, integration tests, manual testing) would improve transparency.

Minor Suggestions:

• Consider adding a section in the PR description summarizing the performance improvements (if any) resulting from the changes.
• Review the use of re.sub in Airforce for filtering. Depending on the complexity of the patterns to be removed, this might be a candidate for optimization using compiled regular expressions.

Overall, this is a high-quality pull request that significantly enhances the project. Addressing the minor suggestions would further improve its clarity and maintainability. I recommend merging after addressing these points.

[github-actions[bot]]

Review of Pull Request: feat(g4f): Major provider updates and new model support

Overview

Thank you for your contribution to the project! This pull request implements significant updates, adding new models and providers, enhancing existing ones, and performing various maintenance tasks. These changes will undoubtedly enhance the functionality of the codebase.

Core Changes

• New Models Added: The addition of models such as o1-preview, o1-mini, llama-3.2-90b, and others is a great enhancement. This allows users access to the latest AI capabilities.
• New Providers Added: Introducing providers like Copilot, LLMPlayground, and AmigoChat expands the availability of tools for users, which is a fantastic addition.
• Provider Improvements: Enhancing conversation handling in DDG and Copilot, along with other fixes and updates to various providers, showcases a commitment to maintaining high-quality interactions.
• Maintenance Tasks: The cleaning up of certain providers, moving them to appropriate directories, and updating mappings reflect a thorough maintenance effort that improves clarity and organization.

Breaking Changes

• Removing outdated models from Liaobots might affect users relying on those models. Adequate documentation and communication regarding this change will be essential to mitigate any issues for current users.

Code Review

The code changes are well-organized, and the implementation seems to follow good practices. The newly added function split_message looks beneficial for breaking down large messages, promoting cleaner and more efficient processing.

Suggestions

1. Documentation: Ensure that documentation is updated for the new models and providers to help users understand how to leverage them effectively.
2. Testing: Consider adding unit tests, if not already present, to cover the newly added models and providers to ensure their correctness and reliability in usage.

Conclusion

Overall, this pull request demonstrates excellent work in enhancing the project significantly. The community will surely benefit from these updates. Thank you once again for your contribution!

Best, g4f copilot

[github-actions[bot]]

# Review for Pull Request: feat(g4f): Major provider updates and new model support

## Summary
Thank you for your contribution to the project! This pull request introduces significant updates to the providers and adds support for several new models, which is a fantastic enhancement for our users.

## Core Changes
- **New Models Added:**
  - `o1-preview`
  - `o1-mini`
  - `llama-3.2-90b`
  - `codellama-34b`
  - `mixtral-7b`
  - `mistral-tiny`
  - `claude-3.5-haiku`
  - `qwen-2.5-72b`
  - `flux-dev`
  - `dalle-3`
  - `recraft-v3`

- **New Providers Added:**
  - `Copilot`
  - `LLMPlayground`
  - `AmigoChat`

- **Base Providers Added:**
  - `Gryphe`
  - `databricks`
  - `anthracite-org`
  - `ai21`
  - `llmplayground.net`

## Provider Improvements
- Enhanced conversation handling for `DDG` and `Copilot`.
- Fixed system prompts in `ChatGptEs` and `Blackbox`.
- Improved message processing in `Airforce`.
- Updated `Liaobots` and `DeepInfra` with the latest models.
- Added support for `Qwen2.5` and `Nemotron` across multiple providers.

## Maintenance
- Moved `PollinationsAI` to `needs_auth/`.
- Moved `AIUncensored` to `not_working/`.
- Removed deprecated `Allyfy` provider.
- Updated model mappings and provider configurations.

## Breaking Changes
- Removed several outdated models from `Liaobots`.

## Code Review
The code changes look well-structured and follow the project's coding standards. The addition of the `split_message` function in `Airforce.py` is a great improvement for handling long messages. The updates in `AmigoChat.py` and other provider files are clear and enhance functionality.

### Suggestions
- Consider adding unit tests for the new models and providers to ensure they work as expected.
- Documentation updates may be necessary to reflect the new models and providers added.

Overall, this is a solid contribution that enhances the

[kqlio67]

@TheFirstNoob I didn't quite grasp the issue and renamed Airforce to LLMPlayground, but after seeing your comments, I reverted back to the original Airforce name to maintain consistency.

I've implemented a straightforward solution by removing the "origin": "https://llmplayground.net" parameter. This resolved the issue with non-working text models.

Regarding splitting the providers, I don't see a practical benefit in maintaining two separate providers since llmplayground.net is essentially just a GUI interface that utilizes api.airforce backend. Given that llmplayground.net's API is fundamentally dependent on api.airforce, the more elegant solution was to simply remove the origin parameter, which has effectively addressed the issue you reported.

[TheFirstNoob]

@kqlio67 Thanks for answer! Let me explain in more detail. The file version from version 0.3.3.4 works with models that are not listed on the site itself, they were listed there earlier, but the functionality works and quite quickly. This list of models is listed above in the bug list and they must be specified in models.py. At the same time, the LLM version, that is, everything after 0.3.3.4, works extremely strangely. On 3.5-turbo and those models that are listed in the bug list, they give me an error that "such a model is not supported" (since we are checking for models that are listed in /models/v1. And all subsequent models, when requested, freeze for me and endlessly wait for a response.

It is for this reason that I suggest separating these two file versions, since there will be greater availability of old and new models. Or a different approach is required.